In the CAIDE 1998 MRI population, lipid-lowering drugs seemed to be protective against more severe WML. However, serum total cholesterol levels were not associated with WML or cortical thickness. Theoretically, hypercholesterolemia-induced atherosclerosis could affect cerebral arterioles, leading to GM and WM changes. Cholesterol levels have also been associated with Alzheimer pathology (Björkhem et al., 2009, Launer et al., 2001. However, previous studies focusing on serum cholesterol levels and GM or WM changes have produced conflicting results (den Heijer et al., 2005a, Koschack et al., 2009, Leritz et al., 2011, Wolf et al., 2004).
Only total serum cholesterol levels were measured at midlife in the CAIDE study.
Guidelines for the prevention and treatment of CHD and ischemic stroke emphasize the importance of LDL and HDL in the development of atherosclerosis (Catapano et al., 2011). The levels of LDL- and HDL-cholesterol will also need to be taken into account when evaluating the effects of cholesterol on brain changes.
6.5 CORONARY HEART DISEASE AND STRUCTURAL BRAIN CHANGES
CHD was associated with decreased cortical thickness and lower total GM volume, and this relation was particularly strong for CHD with a longer duration (at least 10 years). A previous cross-sectional study in male twins linked CHD to lower total GM volume (DeCarli et al., 1999b), but other studies have failed to detect this kind of association (Geerlings et al., 2010, Koschack and Irle, 2005, Manolio et al., 1994). The decline in brain volume observed in patients with CHD has been previously reported in regions related to AD such as temporal lobe, posterior cingulate and precuneus (Almeida et al., 2008, Koschack and Irle, 2005). In Study III, a similar
65
association between CHD and thinner cortex in parietal lobe structures was observed. Other regions related to cognitive functions, such as fusiform gyrus (Han et al., 2012, Tijms et al., 2013) and prefrontal cortex (Funahashi, 2001), also had a lower thickness in the CHD group.
Individuals with CHD often have other co-morbidities such as hypertension, hypercholesterolemia or other cardiovascular diseases. Similarly to CHD, hypertension-related brain changes are more frequently observed in the frontal lobe (Beauchet et al., 2013, Maillard et al., 2012). Cortical thinning and total GM volume loss were most evident in CAIDE participants with both CHD and midlife hypertension, pointing to a combined effect of these two factors. Furthermore, subjects with CHD and a declining SBP displayed more pronounced cortical thinning and lower total GM volume, a pattern similar to that observed in Study II.
Declining BP values could also be a marker of left ventricular dysfunction, which can occur as a consequence of myocardial ischemia. Only four subjects with CHD had been diagnosed with heart failure in the CAIDE 2005-2008 MRI population, and adjusting the analyses for HF did not change the results. No quantitative measures of ventricular function (e.g. ejection fraction) were available, so milder forms of heart failure could not be identified. After adjusting for midlife hypertension in the analyses of the relationship between CHD and cortical thickness, some regions still showed significant associations, suggesting that the effects of CHD on the brain may be partly independent of BP.
Other cardiac diseases such as heart failure and atrial fibrillation may also increase the risk of dementia (Bunch et al., 2010, Miyasaka et al., 2007, Qiu et al., 2006).
Studies of heart failure, atrial fibrillation and structural brain changes have reported less conflicting results in comparison to studies focusing on the relation between CHD and brain changes (Almeida et al., 2012, de Leeuw et al., 2000, Stefansdottir et al., 2013, Vogels et al., 2007). CHD and hypertension are both well known risk factors for heart failure and atrial fibrillation, which may in turn contribute to brain atrophy (Roman, 2004, Stefansdottir et al., 2013). These associations could not be studied in the CAIDE MRI populations due to the small number of participants with these distinct disorders (4 with heart failure and 5 with atrial fibrillation).
CHD diagnoses in the Finnish Hospital Discharge Register represent CHD severe enough to require hospitalization and there for milder forms of CHD could not be identified. In addition, people with severe CHD are less likely to survive to older ages, and thus the findings from Study III may actually underestimate the effects of CHD on the brain.
APOE 4 has been associated with higher risk of CHD (Bennet et al., 2007) and also with more severe coronary atherosclerosis (Kosunen et al., 1995). However, the APOE genotype did not influence the association between CHD and MRI measurements in Study III, possibly due to lack of statistical power.
66
6.6 CAIDE DEMENTIA RISK SCORE AND STRUCTURAL BRAIN CHANGES
CAIDE Dementia Risk Score is a validated tool for estimating the 20 to 40 year dementia risk in the general population based on a midlife risk profile, but it was previously unclear if its dementia prediction ability relies more on cerebrovascular than on neurodegenerative pathology. A higher CAIDE Dementia Risk Score in midlife was associated with more severe WML 20 to 30 years later. An association with more pronounced MTA was observed in the population with longer follow-up times. Individuals with higher CAIDE Dementia Risk Score tended to have thinner cortex in several regions (e.g. posterior cingulate gyrus, temporal lobe, insular cortex), but these results did not remain after correction for multiple comparisons.
There was no relation with total GM volume.
Since the risk score is based largely on vascular factors previously linked to WM and GM changes (Carmelli et al., 1999, de Leeuw et al., 1999, Gustafson et al., 2004, Gustafson et al., 2004, Korf et al., 2004, Soreca et al., 2009), it is not surprising that both more severe WML and MTA were found in individuals with higher midlife CAIDE Dementia Risk Score. Cerebrovascular lesions can be present in AD, especially at older ages, and they can lower the threshold for dementia (Neuropathology Group. Medical Research Council Cognitive Function and Aging Study, 2001, Snowdon et al., 1997). Hippocampal atrophy is a characteristic feature of AD (Burton et al., 2009), but it can also be present in VaD (Jack et al., 2002, Shiino et al., 2012, Yin et al., 2014). Previously, interactions between vascular and Alzheimer pathology have been described (Niwa et al., 2000, Thal et al., 2008).
A higher dementia risk score in midlife was associated with both visually rated WML in the first CAIDE examination and WML volume in the second re-examination. Interestingly, more pronounced MTA was seen only in participants with higher risk scores in midlife and with longer follow-up times (i.e. second CAIDE re-examination). MTA is one of the diagnostic and early markers of AD (Dubois et al., 2007, Jack et al., 2013), and it has been pathologically characterized by an elevated NFT burden (Braak and Braak, 1991, Polvikoski et al., 2010). While WML may be common in the general population even at younger ages (Launer, 2004), MTA may take longer time to develop. The associations between midlife CAIDE Dementia Risk Score and cortical thickness in the second re-examination were not significant after correction for multiple comparisons, but it is worth noting that the principal identified regions (posterior cingulate gyrus, temporal lobe, and insular cortex) follow previously reported brain atrophy patterns in AD (Lerch et al., 2005).
In the initial publication, adding APOE 4 carrier status to the CAIDE Dementia Risk Score did not significantly improve the scale’s ability to predict dementia (Kivipelto et al., 2006). In Study IV, the risk score version including APOE was related to more severe visually rated WML in the first CAIDE re-examination, but not to the other MRI outcomes in the first or second re-examination. Since information about APOE genotype was not available for all participants, this may be
67
at least partly due to limited statistical power. Some previously postulated midlife risk factors for dementia such as diabetes mellitus, depressed mood, head trauma, central obesity, lung function or smoking are not included in the original CAIDE Dementia Risk Score, but in a recent study adding these factors into the analysis did not increase dementia prediction accuracy (Exalto et al., 2013).
6.7 METHODOLOGICAL CONSIDERATIONS
The four studies included in this thesis are based on data from the longitudinal population-based CAIDE study. CAIDE is one of the few studies with available detailed health-related information already at midlife, two re-examinations with total follow-up time of up to 30 years, and which has been specifically designed to investigate risk factors for dementia and AD.
One of the main limitations of Studies I-IV is the relatively small sample size, reducing their statistical power. This may have led to an underestimation or even a failure to observe the associations (type II error) between vascular risk factors/conditions and the MRI measurements. Small sample size also limited the possibility to do comprehensive stratified analyses based on participants’ cognitive status. The CAIDE MRI populations include selected individuals who participated in the first or second re-examination. In the CAIDE 1998 MRI population (39 individuals with dementia, 31 with MCI and 42 controls, age- and sex-matched), data weighting was used to achieve representativeness with the original CAIDE sample. This was not possible in the 2005-2008 MRI population (37 dementia, 70 MCI and 6 controls). Instead, the analyses focused on participants at risk of dementia, and those already with dementia were excluded. Including a high proportion of people with dementia would have affected the results for two main reasons: manifest dementia involves rather pronounced brain changes, with the risk of misidentification/overestimation of associations with vascular factors; and pronounced brain abnormalities can also affect the quality of the automatic segmentation of MRIs. The 69 participants at risk of dementia in the 2005-2008 CAIDE MRI population were not significantly different from the rest of the individuals in the original CAIDE population with respect to age at baseline (p=0.3), gender (p=0.8), education (p=0.2), midlife SBP (p=0.2) or DBP (p=0.6), total cholesterol (p=0.09), BMI (p=0.8), physical activity (p=0.4), APOE genotype (p=0.7), or CHD diagnosed during the study (p=0.4).
Although MRI was performed in both re-examinations, only 18 subjects had MRI from both time points. It was thus not possible to analyze changes in MRI measurements over time in relation to vascular factors. The MRI acquisition parameters and scanners were also different in the first and second re-examinations, limiting image analysis in some cases (e.g. WML volume and cortical thickness could not be measured reliably on some images from the first re-examination).
68
Non-participation and survival bias may have also influenced the results.
Individuals in the original CAIDE target population who died or did not participate in re-examinations had poorer health status and higher vascular risk (e.g. higher BP, BMI and cholesterol) compared to survivors/participants (Kulmala et al., 2014). This may have led to an underestimation of the effects of vascular risk factors on the brain.
Cognitive testing and MRI were not available at the baseline (midlife) visit.
Alzheimer and cerebrovascular pathologies can start to develop long before any dementia diagnosis, and the possibility of reverse causality needs to be taken into account. However, if there were some individuals exhibiting the early stages of dementia (and more pronounced brain pathology) already at baseline, they would have been unlikely to survive and participate in re-examinations. Potential GM atrophy or WML in participants who later on developed dementia would nor have been anticipated to be severe enough to have a major influence on BP levels at midlife (Dickerson et al., 2011, Jack et al., 2013).
The CAIDE study provided a large amount of information about vascular and lifestyle-related factors, and analyses in Studies I-IV took into account several potential confounders and effect mediators. This is particularly important as vascular risk factors may affect the brain through shared pathways, and they can also interact with each other. However, the possibility of residual confounding cannot be fully excluded (e.g. less severe comorbid cardiovascular conditions who did not require hospitalization).
69
7 Conclusions
Based on the findings from the present set of studies, the following conclusions can be drawn:
1) Midlife hypertension was associated with an increased risk of more severe WML and lower cortical thickness 20 to 30 years later. Individuals with longstanding hypertension and those who developed hypertension at older ages also had an increased risk of WML. A decline in blood pressure from midlife to late-life was observed in subjects with thinner cortex in brain areas involved in blood pressure regulation (e.g. insular cortex).
2) The presence of midlife overweight and obesity were related to an increased risk of more severe WML 20 years later. Elevated BMI from midlife to late-life was associated with WML in late-life.
3) Although the serum total cholesterol level was not related to brain MRI measurements, lipid-lowering treatment seemed to exert a protective effect against WML.
4) Lower total GM volume and reduced cortical thickness in several brain regions were found in subjects with coronary heart disease, particularly in those with a longer disease duration. This association was influenced by midlife blood pressure levels and changes in blood pressure over time.
5) Higher midlife CAIDE Dementia Risk Score was associated with more severe WML and MTA 20 to 30 years later.
The results of this project emphasize that vascular risk factors and conditions existing from midlife to older ages can influence the structural brain changes detected later with MRI. A longer exposure time to such factors is particularly detrimental. A validated, easy to use risk score for estimating dementia risk based on vascular factors can also point to an increased risk for cerebrovascular and neurodegenerative changes, and could be useful for identifying at-risk individuals who may benefit most from preventive interventions.
70
71
8 Future Perspectives
Evidence from observational studies supports the hypothesis that there is an association between vascular risk factors already in midlife and late-life cognitive impairment, creating a window of opportunity for prevention. Interestingly, recent studies have hinted that the incidence of dementia may be indeed declining (Matthews et al., 2013, Qiu et al., 2013, Schrijvers et al., 2012). One possible explanation for this phenomenon is that there have been changes in cardiovascular risk factors since the 1960s-1970s, i.e. decreasing prevalence of hypertension, hyperlipidemia or smoking. However, both overweight and DM are becoming more common in middle-aged and older populations (Fielding et al., 2013, Finucane et al., 2011, Luchsinger, 2010, Vartiainen et al., 2010). The midlife assessment of CAIDE participants took place during a time when the levels of many vascular risk factors were generally high throughout Eastern Finland (Puska et al., 1979). New epidemiological studies are needed to confirm the trend of declining dementia incidence, and also to investigate vascular risk factors in the new generations of older people, since it is clear that conditions in the population can change significantly during the time when long-term follow-up studies are on-going. It also remains to be determined whether changes in dementia incidence are accompanied by changes in the type and severity of brain pathology.
Although observational studies have indicated that treatment of vascular factors (e.g. antihypertensive treatment, lipid-lowering medication) may decrease the risk of dementia and slow cognitive decline (Chang-Quan et al., 2011, Deschaintre et al., 2009, Luchsinger et al., 2007, Rockwood et al., 2002), these promising findings have not been easily translated into successful dementia prevention in randomized controlled trials (RCT) (Ligthart et al., 2010, Richard et al., 2012b). However, these RCTs were often add-on studies in trials focusing on decreasing cardiovascular mortality and preventing cardio- or cerebrovascular events. They also tended to include younger populations (<70 years), which resulted in a relatively low incidence of dementia and cognitive impairment, and thus they had a limited power to detect significant treatment effects (Ligthart et al., 2010). These methodological issues will need to be addressed in future prevention RCTs.
Future RCTs focusing on prevention of cognitive decline would be advised to include biomarkers for dementia-related diseases (e.g. MRI, PET or CSF markers) in order to better assess the overall effects of the intervention. Antihypertensive treatment has been postulated to exert a beneficial effect on WML progression in observational studies (Godin et al., 2011, Verhaaren et al., 2013), but no change in GM atrophy was seen in a 1-year follow-up study despite successful antihypertensive treatment (Jennings et al., 2011). In a 2-year RCT, statin therapy slowed the progression of WML in the group with severe WML already at baseline
72
compared to the control group (Mok et al., 2009), but these results have not been confirmed in other RCTs (ten Dam et al., 2005).
Late-life cognitive impairment is a heterogeneous condition and focusing only on treating a single risk factor may not be enough. RCTs targeting several risk factors simultaneously may be more likely to represent effective prevention strategies. Such multi-domain RCTs are already ongoing in several European countries, e.g. the European Dementia Prevention Initiative (EDPI, http://www.edpi.org) (Dehnel 2013, Richard et al., 2012). The EDPI includes three RCTs already in progress: in Finland, the Finnish Geriatric Intervention study to prevent cognitive impairment and disability (FINGER) (Kivipelto et al., 2013); in the Netherlands, the Prevention of Dementia by Intensive Vascular Care (preDIVA) trial (Richard et al., 2009); and in France, the Multidomain Alzheimer Prevention Trial (MAPT) (Carrié et al., 2012).
Cognitive functioning or dementia are the primary outcomes in these trials, and they also include several neuroimaging exploratory outcomes: basic structural MRI modalities, and additionally DTI, FDG-PET for brain glucose metabolism, and PiB-PET for brain amyloid are used in sub-groups in FINGER and MAPT. FINGER, MAPT and preDIVA have intervention periods of 2, 3 and 6 years, respectively, and planned follow-up periods of 7, 5 and 6 years (Richard et al., 2012a). Since brain changes on structural MRI and FDG-PET can be seen already years before AD/dementia (Jack et al., 2013), the results concerning the effects of vascular and lifestyle preventive interventions on these biomarkers will be of high interest.
A fourth trial, Healthy Aging Through Internet Counselling in the Elderly (HATICE) (http://www.hatice.eu/), focusing on management of vascular risk factors and conditions, is planned to start in 2015 in the Netherlands, Finland and France.
73
9 References
Albert, M.S., DeKosky, S.T., Dickson, D., Dubois, B., Feldman, H.H., Fox, N.C., Gamst, A., Holtzman, D.M., Jagust, W.J., Petersen, R.C., Snyder, P.J., Carrillo, M.C., Thies, B., Phelps, C.H., 2011. The diagnosis of mild cognitive impairment due to Alzheimer's disease: recommendations from the National Institute on Aging-Alzheimer's Association workgroups on diagnostic guidelines for Aging-Alzheimer's disease. Alzheimers Dement. 7, 270-279.
Almeida, O.P., Garrido, G.J., Beer, C., Lautenschlager, N.T., Arnolda, L., Flicker, L., 2012. Cognitive and brain changes associated with ischaemic heart disease and heart failure. Eur.Heart J. 33, 1769-1776.
Almeida, O.P., Garrido, G.J., Beer, C., Lautenschlager, N.T., Arnolda, L., Lenzo, N.P., Campbell, A., Flicker, L., 2008. Coronary heart disease is associated with regional grey matter volume loss: implications for cognitive function and behaviour.
Intern.Med.J. 38, 599-606.
American Psychiatric Association, 2013. Diagnostic and Statistical Manual of Mental Disorders, the fifth revision.
American Psychiatric Association, 1994. Diagnostic and statistical manual of mental disorders 4th edn.
Anbeek, P., Vincken, K.L., van Osch, M.J., Bisschops, R.H., van der Grond, J., 2004.
Probabilistic segmentation of white matter lesions in MR imaging. Neuroimage. 21, 1037-1044.
Appelman, A.P., Exalto, L.G., van der Graaf, Y., Biessels, G.J., Mali, W.P., Geerlings, M.I., 2009. White matter lesions and brain atrophy: more than shared risk factors? A systematic review. Cerebrovasc.Dis. 28, 227-242.
Aronson, M.K., Ooi, W.L., Morgenstern, H., Hafner, A., Masur, D., Crystal, H., Frishman, W.H., Fisher, D., Katzman, R., 1990. Women, myocardial infarction, and dementia in the very old. Neurology. 40, 1102-1106.
Ashburner, J., Friston, K.J., 2000. Voxel-based morphometry--the methods.
Neuroimage. 11, 805-821.
Atti, A.R., Palmer, K., Volpato, S., Winblad, B., De Ronchi, D., Fratiglioni, L., 2008.
Late-life body mass index and dementia incidence: nine-year follow-up data from the Kungsholmen Project. J.Am.Geriatr.Soc. 56, 111-116.
74
Baezner, H., Blahak, C., Poggesi, A., Pantoni, L., Inzitari, D., Chabriat, H., Erkinjuntti, T., Fazekas, F., Ferro, J.M., Langhorne, P., O'Brien, J., Scheltens, P., Visser, M.C., Wahlund, L.O., Waldemar, G., Wallin, A., Hennerici, M.G., LADIS Study Group, 2008. Association of gait and balance disorders with age-related white matter changes: the LADIS study. Neurology. 70, 935-942.
Baloh, R.W., Ying, S.H., Jacobson, K.M., 2003. A longitudinal study of gait and balance dysfunction in normal older people. Arch.Neurol. 60, 835-839.
Banach, M., Aronow, W.S., 2012. Blood pressure j-curve: current concepts.
Curr.Hypertens.Rep. 14, 556-566.
Barnes, D.E., Covinsky, K.E., Whitmer, R.A., Kuller, L.H., Lopez, O.L., Yaffe, K., 2010. Commentary on "Developing a national strategy to prevent dementia: Leon Thal Symposium 2009." Dementia risk indices: A framework for identifying individuals with a high dementia risk. Alzheimers Dement. 6, 138-141.
Barnes, D.E., Covinsky, K.E., Whitmer, R.A., Kuller, L.H., Lopez, O.L., Yaffe, K., 2009. Predicting risk of dementia in older adults: The late-life dementia risk index.
Neurology. 73, 173-179.
Barnes, D.E., Yaffe, K., 2011. The projected effect of risk factor reduction on Alzheimer's disease prevalence. Lancet Neurol. 10, 819-828.
Bateman, R.J., Xiong, C., Benzinger, T.L., Fagan, A.M., Goate, A., Fox, N.C., Marcus, D.S., Cairns, N.J., Xie, X., Blazey, T.M., Holtzman, D.M., Santacruz, A., Buckles, V.,
Bateman, R.J., Xiong, C., Benzinger, T.L., Fagan, A.M., Goate, A., Fox, N.C., Marcus, D.S., Cairns, N.J., Xie, X., Blazey, T.M., Holtzman, D.M., Santacruz, A., Buckles, V.,